[llvm] c9c18e5 - [AMDGPU] Disallow dpp combining for dpp instructions without Src2 operand (when Src2 is required)

[via llvm-commits]

Author: vpykhtin
Date: 2019-10-25T21:30:37+03:00
New Revision: c9c18e5a319420d9caea66836818f7e9f8fc42de

URL: https://github.com/llvm/llvm-project/commit/c9c18e5a319420d9caea66836818f7e9f8fc42de
DIFF: https://github.com/llvm/llvm-project/commit/c9c18e5a319420d9caea66836818f7e9f8fc42de.diff

LOG: [AMDGPU] Disallow dpp combining for dpp instructions without Src2 operand (when Src2 is required)

Differential revision: https://reviews.llvm.org/D69430

Added: 
    

Modified: 
    llvm/lib/Target/AMDGPU/GCNDPPCombine.cpp
    llvm/test/CodeGen/AMDGPU/dpp_combine.mir

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Target/AMDGPU/GCNDPPCombine.cpp b/llvm/lib/Target/AMDGPU/GCNDPPCombine.cpp
index 98678873e37c..edf2b5f62b95 100644
--- a/llvm/lib/Target/AMDGPU/GCNDPPCombine.cpp
+++ b/llvm/lib/Target/AMDGPU/GCNDPPCombine.cpp
@@ -235,7 +235,8 @@ MachineInstr *GCNDPPCombine::createDPPInst(MachineInstr &OrigMI,
     }
 
     if (auto *Src2 = TII->getNamedOperand(OrigMI, AMDGPU::OpName::src2)) {
-      if (!TII->isOperandLegal(*DPPInst.getInstr(), NumOperands, Src2)) {
+      if (!TII->getNamedOperand(*DPPInst.getInstr(), AMDGPU::OpName::src2) ||
+          !TII->isOperandLegal(*DPPInst.getInstr(), NumOperands, Src2)) {
         LLVM_DEBUG(dbgs() << "  failed: src2 is illegal\n");
         Fail = true;
         break;

diff  --git a/llvm/test/CodeGen/AMDGPU/dpp_combine.mir b/llvm/test/CodeGen/AMDGPU/dpp_combine.mir
index e0e3c352fd20..0c4d0e0e2384 100644
--- a/llvm/test/CodeGen/AMDGPU/dpp_combine.mir
+++ b/llvm/test/CodeGen/AMDGPU/dpp_combine.mir
@@ -794,3 +794,19 @@ body: |
     %4:vgpr_32 = V_ADD_I32_e32 %2.sub0, undef %3:vgpr_32, implicit-def $vcc, implicit $exec
     %5:vgpr_32, dead %6:sreg_64_xexec = V_ADDC_U32_e64 1, %2.sub1, undef $vcc, 0, implicit $exec
 ...
+
+# GCN-LABEL: name: dont_combine_cndmask_with_src2
+# GCN: %5:vgpr_32 = V_CNDMASK_B32_e64 0, %3, 0, %1, %4, implicit $exec
+name: dont_combine_cndmask_with_src2
+tracksRegLiveness: true
+body: |
+  bb.0:
+    liveins: $vgpr0, $vgpr1
+    %0:vgpr_32 = COPY $vgpr0
+    %1:vgpr_32 = COPY $vgpr1
+    %2:vgpr_32 = IMPLICIT_DEF
+
+    %3:vgpr_32 = V_MOV_B32_dpp %2, %0, 1, 15, 15, 1, implicit $exec
+    %4:sreg_64_xexec = IMPLICIT_DEF
+    %5:vgpr_32 = V_CNDMASK_B32_e64 0, %3, 0, %1, %4, implicit $exec
+...